[Pkg-ganeti-devel] Some questions about packaging

[Leonardo Rodrigues de Mello]

Considerations:
I agree with the issues related to orig tar.gz and ITP.
I believe the name would be ganeti-os-debian-etch.
Because it is one os script for ganeti, it isn't one instance.
The instance will be created using this script.

And i don't see why not use the version of this package to describe the api.

That way we could use the API number as major package version.
ex: API 4, version xyz, package release 2
ganeti-os-debian-etch version 4.xyz-2

We can use conflict on ganeti with ganeti-os-debian-etch  (<  4)  for
example.

We can use Provide to provide ganeti-os virtual package.
That way future os scripts also provide this package.

2008/1/18, Iustin Pop <iusty at k1024.org>:
>
> On Fri, Jan 18, 2008 at 04:52:12PM +0100, Guido Trotter wrote:
> > On Fri, Jan 18, 2008 at 03:29:58PM +0100, Iustin Pop wrote:
> >
> > Ciao,
> >
> > > First, using svn-buildpackage we need to have the orig tarball as a
> > > tar.gz and not just as a simple .tar. This is not nice to fix (we
> > > provide a different tarball than upstream), and from next upstream
> > > version we'll switch to .tar.gz; it's more consistent this way.
> > >
> >
> > That's good! :) :) Well, for now changing the tarball is not too bad,
> I've done
> > worse things to "orig" tarballs! :) :)
> he he :)
>
> > > Second, I think there should be an ITP first, right? I'll open one. Do
> > > we agree that the name should be ganeti-instance-debian-etch?
> > >
> >
> > LGTM
> ok, will open the ITP.
>
> > > Third, lintian/linda warn me that an arch-independent package installs
> > > something under /usr/lib. I don't know a nice workaround to this,
> except
> > > by extending ganeti's search patch to cover also /usr/share/ganeti/os.
> > >
> >
> > That's probably the right way to go...
> ok, I'll add that to the ganeti pkg.
>
> > > Also, should we make some relationship between what os-api is provided
> > > by ganeti and what an OS provides? For example, ganeti recommends
> > > ganeti-os-api version 4 (a virtual package);
> ganeti-instance-debian-etch
> > > has a Provides: ganeti-os-api.
> > >
> >
> > I don't think there can be versioned provides.. Can there?
>
> Good point. It seems not:
> "So, a Provides field may not contain version numbers, and the version
> number of the concrete package which provides a particular virtual
> package will not be looked at when considering a dependency on or
> conflict with the virtual package name."
>
> Therefore, we could have the virtual names "ganeti-os-api-4", next time
> "ganeti-os-api-5", etc.
>
> But I'm also not sure about this section:
> "All packages should use virtual package names where appropriate, and
> arrange to create new ones if necessary. They should not use virtual
> package names (except privately, amongst a cooperating group of
> packages) unless they have been agreed upon and appear in the list of
> virtual package names. (See also Virtual packages - Provides, Section
> 7.4)".
>
> Could we say here that the ganeti packages are using these names
> "privately"?
>
> iustin
>
> _______________________________________________
> Pkg-ganeti-devel mailing list
> Pkg-ganeti-devel at lists.alioth.debian.org
> http://lists.alioth.debian.org/mailman/listinfo/pkg-ganeti-devel
>



-- 
Leonardo Rodrigues de Mello
jabber: l at lmello.eu.org
-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://lists.alioth.debian.org/pipermail/pkg-ganeti-devel/attachments/20080118/2a5ae97b/attachment.htm